Better citations needed: a big upgrade to the PainSci bibliography

In 2015, a huge bibliography and “good footnotes” still set PainScience apart.1My footnotes contain either extra commentary and whimsical asides, or citations to science and other sources, like this:

Woolf CJ. Central sensitization: Implications for the diagnosis and treatment of pain. Pain. 2010 Oct;152(2 Suppl):S2–15. PubMed 20961685 ❐ PainSci Bibliography 54851 ❐ And I’ve just upgraded them.

Bibliographic data does not play nicely with modern publishing technology. There’s lots of software for wrangling references on your PC, but it’s still almost impossible to integrate them (efficiently) into blogs and websites. It still has to mostly be done “manually.”

I started investing in a good system to solve this problem in 2005, but it had to be custom job — pure original programming. The result has been a quietly awesome and esoteric boon to my publishing business ever since. My referencing gets noticed by many visitors to the site, because there’s nothing else quite like it anywhere else online. Footnotes in long-form online writing are sparse and spartan.

But it was time for some improvements. Got stay ahead of the curve. This is about what I did, and some behind-the-scenes details.

A craaaazy amount of work

Pulling this off tooks weeks of work spread out over months, but almost everything is better:

• More information! I revised the summaries of over 400 most-cited papers (out of about 2500). Imagine going back and editing a couple year’s worth of regular science blogging. 100 of the best examples are linked below.
• Faster, nicer! Those revised summaries load up about 10× faster, and they’re easier on the eyes.
• Organized! The index to the entire bibliography is also much faster and it sucks much less.
• More formal! I converted my referencing format to the “Vancouver system,” the standard used by most medical journals.

That last one was particularly challenging and interesting. Why “Vancouver”? My town?

The old way: dumbed-down references

PainScience isn’t a medical journal, and all of my readers in the early days were “just folks,” so why put on airs? I made my references simple: just the essential information, uncluttered with stuff like 82(123):987-93. (But always with a link to those details, so that a reference could be checked up on — that’s very important, even for just folks.)

It was idealistic and modern. I was declaring what I thought web referencing should be like.

Plus, it is really hard to mass produce the formatting of detailed bibliographic data. So I didn’t.

The new way: more formal = more serious

It’s hard to get taken seriously when you don’t act serious. Regular readers aren’t bothered by formality and detail, not in a footnote, but special visitors — proper experts — definitely notice if I don’t speak their referencing language. Those influential readers won’t know that I’m being clever and innovative and democratic.

“They’ll just think you don’t know how to reference properly,” my wife said about five years ago.

“And they’ll be right,” I said. “I don’t really have a clue. It looks nasty.”

“Time to learn, maybe.”

“Maybe. I should get started on the procrastinating, at least.”

And then I started getting email from people who agreed with my wife. My writing was getting noticed. My website was getting big. Getting referencing right mattered more with each passing year.

[better citations needed]

The Vancouver system of referencing

About six months ago I Googled “standard medical journal referencing format.” After all these years, I wasn’t even sure if there was such a standard. That’s when I discovered “the Vancouver system,” and it kicked off months of work. I had to re-tool the footnote factory & re-train all the bibliography gnomes. (Weirdly, I felt much more comfortable diving into this Sysyphean chore simply because the new standard was named after where I lived.)

So, why Vancouver?

In 1978, editors of medical journals from around the world met in the city of Vancouver, probably close to where I live, and thrashed out the unbelievably numerous details. It was so difficult and tedious that they named the standard after the city they were trapped in. Their work is still the standard today, and it is heavily documented.

I encourage you to click that link and scroll for a while and behold my nightmare. The Vancouver system is about as user-friendly as a swarm of cranky wasps.

The devil is in the details

All my references are generated from a bibliographic database in the fairly exotic BibTeX data format. Every footnote is lovingly crafted by software — essential for mass production. I had to reprogram that software to speak “Vancouver style.”

In programming, an “edge case” is something that requires special handling. You might have 50 lines of code that do something with 95% of your data, and then another 50 lines of code just to cope with a few awkward exceptions in the last 5%. Bibliographic data seems like it’s all edge. It is inherently rotten with the influences of other languages, the strange ways of databases, and oddball conventions grandfathered in from Ye Olde Dayes. My own data collection methods varied over the years: what I kept and how I kept it has come back to haunt me.

I lost entire days to minutiae like accented characters, abbreviations, pagination punctuation, title casing methods, and so on. I nearly lost my mind trying to either hammer my data into the necessary shape, or write clever subroutines that would do the equivalent, or both.

And fairly often I just flat-out disagreed with the Vancouver system. It has some major flaws. My goal was to look like I know the standard — and I do now — but I’m not going to follow it off a typographic cliff.

Old versus new

So here’s an old citation …

Vibe-Fersum et al. Efficacy of classification-based cognitive functional therapy in patients with non-specific chronic low back pain: A randomized controlled trial. European Journal of Pain. 2013. PubMed 23208945 ❐

There’s only one author listed, the journal spelled out instead of abbreviated, and there’s no date, volume, and issue data. Here’s the same paper cited Vancouver style …

Vibe-Fersum K, O’Sullivan P, Skouen JS, Smith A, Kvåle A. Efficacy of classification-based cognitive functional therapy in patients with non-specific chronic low back pain: A randomized controlled trial. Eur J Pain. 2013 Jul;17(6):916–28. PubMed 23208945 ❐

Okay, admittedly it’s not a very exciting difference when all is said and done. 😃 It’s getting thousands of those to look right automatically that’s the impressive part.

100 good citation examples

There’s about 2500 items in the bibliography these days. 400 or so got some extra care and editing during the big upgrade. The 100 best and most interesting are listed here. (I can generate lists like this very easily, one of the superpowers of the system: I just tell the database to generate a score for each record by awarding points for things like how recent is, study quality, summary length, and keywords like “fun” and “odd” and “classic” and “good news.”)

1. A critical evaluation of the trigger point phenomenon
2. Spinal manipulation no better for back pain than placebos
3. Central sensitization
4. Functional Movement Screen unreliable
5. Adverse events and cervical manipulation for neck pain
6. Promising trial of cognitive functional therapy for low back pain
7. Massage therapy attenuates inflammatory signaling after exercise-induced muscle damage
8. Forefoot runners have fewer injuries, but causality unclear
9. Neck strength can reduce chronic neck pain long-term
10. Regular Swedish versus “tensegrity-based” massage
11. Location of back and neck pain could not be detected by feel
12. More than 20% of manual therapy treatments do some harm
13. Cellular response to simulated myofascial release
14. Worn out shoes do change the biomechanics of running, but not much
15. Small, flawed trial of foam rolling shows 8% ROM increase
16. Comparison of 2 types of massage for chronic low back pain
17. Trial of therapeutic massage for neck pain
18. Brief, intense muscular training for cardiovascular fitness
19. Patellar maltracking in patellofemoral pain with patella alta
20. Why Most Published Research Findings Are False
21. Curcumin “likely” reduces muscle soreness after exercise
22. Thigh and hip exercises effective for patellofemoral pain
23. Increased trapezius pain sensitivity is not associated with increased tissue hardness
24. Regular hamstring stretching increased range of motion
25. The greatest hits of back pain science are a disappointment
26. Massage therapy probably helps patients with bone cancer
27. Flexibility gains due to changes in sensation, not muscle length
28. Safe but useless for knee arthritis: glucosamine, chondroitin sulphate, and celecoxib
29. The iliotibial band is uniformly, firmly attached to the femur
30. Intense, brief workouts almost as effective as time-consuming cardio
31. Trial of glucosamine for low back pain finds no therapeutic effect
32. Nerve root impingement fairly rare, barely more common in car accident victims
33. A fascinating landmark study of placebo surgery for knee osteoarthritis
34. The hazards of NSAIDs, especially diclofenac
35. Massage vs minimal exercise for poor circulation
36. Education, not core exercise, reduces back pain incidence in soldiers
37. Trigger points are acidic and contain pain-causing metabolites
38. 8 weeks of core strengthening, coordination exercise for chronic low back pain
39. Do strong quadriceps help patellofemoral pain?
40. Surprisingly effective back pain injection: intradiscal methylene blue
41. Quality of online sports medicine information “highly variable”
42. Only quantity of exercise for back pain produces better results
43. Failed trial of vertebroplasty for compression fractures
44. Deyo and Weinstein’s 2001 low back pain tutorial
45. Regular, moderate exercise boosts makes neutrophils busier for longer
46. Yoga, stretching equally and slightly effective for back pain
47. Stress fractures: it’s not how hard you hit the ground, but how fast you hit it
48. Intravascular danger signals guide neutrophils to sites of sterile inflammation
49. Current evidence does not support Botox for trigger points
50. Online tutorials for chronic pain reduced pain, anxiety, disability
51. Prebiotics reduces waking cortisol response
52. Functional implications of the Q-angle in the patellofemoral joint
53. Botox for trigger points, update
54. Strong criticism of “more is better” strength training
55. Chiropractic subluxation is still “unsupported speculation”
56. Special core strengthening prevents no more injuries than ordinary sit-ups
57. Both heat and cold for back and neck strain mildly beneficial
58. Massage impairs post exercise muscle blood flow and lactic acid removal
59. General practitioners do not follow guidelines for low back pain care
60. A review of low quality evidence about exercise for neck pain
61. Chiropractic identity, role and future: survey
62. Is hip strength a risk factor for patellofemoral pain?
63. Regular physical activity prevents chronic pain
64. Stretching and heart rate variability in inflexible subjects
65. Smoking associated with low back pain, intervertebral disc disease
66. Cramps caused by effort, not dehydration and electrolyte shortage
67. Recent injury had no effect on FMS scores
68. Asymmetry of psoas and quadratus lumborum unrelated to injury
69. No clear benefit to muscle relaxants for acute neck strain
70. The science of trigger point diagnosis is a confusing mess so far
71. Myoglobin in plasma after trigger point massage
72. Cherries for soreness? Well, weakness at least
73. What causes the burn in intense muscle effort?
74. Dry needling for myofascial pain, review
75. A disturbing and typical example of sloppy modern acupuncture research
76. Promising results from athroscopic surgery for IT band syndrome
77. Does long-distance running lead to cartilage damage? An MRI study
78. Underwhelming: spinal adjustment and massage for back pain, neck pain
79. The effect of leg length on back pain: a classic test
80. Is exercise effective, or just efficacious?
81. Lumbosacral transition vertebra prevalence, significance
82. The Conundrum of Calcaneal Spurs
83. A major, comprehensive report on treatments for knee arthritis
84. Magnetic resonance imaging in follow-up assessment of sciatica
85. Minor benefits of pilates for chronic low back pain
86. Tuning fork, ultrasound diagnosis of stress fractures is unreliable
87. Hamstring flexibility cannot predict lumbar joint use in reaching
88. Women adapt effectively and minimally to wearing high heels
89. Acupuncture for back pain, a poor quality trial
90. Can trigger point therapy improve restricted ankle joint motion?
91. Review: patients are “highly satisfied” with physical therapy
92. Extremely thorough, valuable review of studies of back pain treatments
93. Stretching, strengthening don’t affect knee and shin injury rates in soldiers in basic training
94. Exercise reduces anterior knee pain risk
95. Prospective Predictors of Patellofemoral Pain Syndrome
96. Kinesio taping in treatment and prevention of sports injuries
97. Icing delays recovery from muscle soreness
98. Deep friction massage to treat tendinopathy: still no evidence
99. Fascia is too tough for mechanical deformation